Cancer mutational processes vary in their association with replication timing and chromatin accessibility

Background Cancer somatic mutations are the product of multiple mutational and repair processes, which are tightly associated with DNA replication. Distinctive patterns of somatic mutations accumulation in tumors, termed mutational signatures, are indicative of processes the tumors underwent. While tumor mutational load is correlated with late replicating regions and spatial genome organization, much is unknown about the association of many different mutational processes and replication timing, and the interplay with chromatin structure remains an open question. Methods We systematically analyzed the mutational landscape of 2,787 WGS tumors from 32 different tumor types separately for early and late replicating regions. We used sequence context normalization and chromatin data to account for sequence and chromatin accessibility differences between early and late replicating regions. Moreover, we expanded the signature analyses to doublet base substitutions and small insertions and deletions by developing an artificial genomes-based approach to account for sequence differences between various genomic regions. Results We revealed the replication timing (RT) association of single base, doublet base and small insertions and deletions mutational signatures. The association is signature specific: some are associated with early or late replication (such as UV-exposure signatures SBS7b and SBS7a, respectively) and others have no association. Most associations exist even after normalizing for genome accessibility. We further developed a focused mutational signature identification approach, which uses RT information to improve signature identification, and found that SBS16, which is biased towards early replication, is strongly associated with better survival rates in liver cancer. Conclusions Our comprehensive analyses enabled a more robust classification of RT association of single base, doublet base and indels signatures. By doing so, we demonstrated a variation in the association with RT, as many mutational processes biased towards either early or late replication timing, and others have an equal RT distribution. These associations were independent from chromatin accessibility in most cases. This work highlights that restricting signatures analyses to concise genomic regions improves identification of signatures, such as SBS16, and demonstrates its clinically relevance as a predictor of improved survival of liver cancer patients.